Extracting Disaster Impacts and Impact Related Locations in Social Media Posts Using Large Language Models

2025-11-23

AI Summary

In large-scale disasters, social media texts often exhibit a “mentioned location ≠ affected location” mismatch, creating spatiotemporal gaps in situational awareness and resource allocation. To address this, we propose the first large language model (LLM) fine-tuning framework specifically designed for disaster-affected location identification. Our method jointly models impact event detection and geographic semantic association, enabling robust parsing of informal expressions, abbreviations, and implicit geographic references. By integrating LLM fine-tuning with fine-grained named entity recognition, we achieve end-to-end impact–location alignment. Evaluated on real-world disaster tweet datasets, our approach achieves F1 scores of 0.69 for impact identification and 0.74 for affected location extraction—substantially outperforming conventional NLP baselines. This work pioneers the systematic application of LLM fine-tuning to disaster location disambiguation, delivering a deployable, real-time, and high-precision solution for disaster sentiment and impact sensing.

Technology Category

• Natural Language Processing: Fact-Checking / Misinformation Detection (NLP Focus)
• Machine Learning: Large Multimodal Models (LMMs)
• Planning, Routing, and Scheduling: Planning with Language Models

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Large language models for search
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

Large-scale disasters can often result in catastrophic consequences on people and infrastructure. Situation awareness about such disaster impacts generated by authoritative data from in-situ sensors, remote sensing imagery, and/or geographic data is often limited due to atmospheric opacity, satellite revisits, and time limitations. This often results in geo-temporal information gaps. In contrast, impact-related social media posts can act as "geo-sensors" during a disaster, where people describe specific impacts and locations. However, not all locations mentioned in disaster-related social media posts relate to an impact. Only the impacted locations are critical for directing resources effectively. e.g., "The death toll from a fire which ripped through the Greek coastal town of #Mati stood at 80, with dozens of people unaccounted for as forensic experts tried to identify victims who were burned alive #Greecefires #AthensFires #Athens #Greece." contains impacted location "Mati" and non-impacted locations "Greece" and "Athens". This research uses Large Language Models (LLMs) to identify all locations, impacts and impacted locations mentioned in disaster-related social media posts. In the process, LLMs are fine-tuned to identify only impacts and impacted locations (as distinct from other, non-impacted locations), including locations mentioned in informal expressions, abbreviations, and short forms. Our fine-tuned model demonstrates efficacy, achieving an F1-score of 0.69 for impact and 0.74 for impacted location extraction, substantially outperforming the pre-trained baseline. These robust results confirm the potential of fine-tuned language models to offer a scalable solution for timely decision-making in resource allocation, situational awareness, and post-disaster recovery planning for responders.

Problem

• Extracts disaster impacts and locations from social media posts
• Distinguishes impacted from non-impacted locations using LLMs
• Enables timely resource allocation and situational awareness

Innovation

• Fine-tuning LLMs to identify disaster impacts and locations
• Distinguishing impacted locations from non-impacted ones in posts
• Extracting informal location expressions for situational awareness